Social and economic losses from natural disasters are potentially staggering. To reduce losses, the immediate and post-crisis response to disasters is important. However, mitigation activities before a natural disaster occurs can be even more effective in reducing losses. Nevertheless, developing and implementing long term mitigation schemes can be difficult, because: decision makers tend to invest in works with clearer short-term benefits; risk attributed to disasters is prone to inaccuracy as disasters are relatively infrequent; the people influencing mitigation activities may have little personal experience to guide their evaluation; and mitigation budgets are always limited, therefore selecting the optimal trade-off of mitigation options can be very difficult. Because of these difficulties, decision support systems (DSS) are advantageous, as they: (1) are transparent and can quantify the expected benefits of mitigation investiture across multiple criteria; (2) assess the likelihood and consequences of natural disasters across multiple criteria; and (3) use formal optimization techniques to find optimal or near-optimal mitigation portfolios. However, DSSs for natural disaster mitigation have tended to focus on disaster preparedness and the immediate and post-crisis response to emergencies. Of those DSSs that have focused on mitigation, none have considered both temporal nonstationarity in climate or land use, and the use of optimization to form mitigation portfolios. Consequently, an integrated natural hazard mitigation DSS is being developed for the state of South Australia that optimizes the choice of mitigation options, through assessing the performance of various policy options in the long term, by evaluating the performance of mitigation options using models across a number of natural hazards in an integrated way, whilst taking account of land use and climate change. This paper introduces the DSS, presents the questions the DSS will help answer, outlines the development approach, and summarizes the optimization and modeling strategies used.